Hinton大师解析神经网络(neural_network)、信念网络(belief_net)、玻尔兹曼机(RBM)
Hinton大师解析神经网络(neural_network)、信念网络(belief_net)、玻尔兹曼机(RBM)Tutorial on:
Deep Belief Nets
GeoffreyHintonCanadianInstitute for Advanced Research&Departmentof Computer ScienceUniversityof Toronto
FOUNDATIONS OF DEEP LEARNING•Why we need to learn generativemodels.
•Why it is hard to learn directedbelief nets.
•Two tricks that make it easy tolearn directed belief nets with an associative memory on top.
•The theoretical justification forthe two tricks.
•
FINE-TUNING TO IMPROVE DISCRIMINATION•Why it works better than purediscriminative training.
•
DEALING WITH DIFFERENT TYPES OF DATA•Three ways to model real values
•How to model bags of words
•How to model high-dimensionalsequential data.
A spectrum of machine learningtasks
TypicalStatistics------------ArtificialIntelligence•Low-dimensionaldata (e.g. less than 100 dimensions)
•
•Lots of noise in the data
•
•There is not much structure in the data, and what structure there is,can be represented by a fairly simple model.
•
•
•Themain problem is distinguishing true structure from noise.
•High-dimensionaldata (e.g. more than 100 dimensions)
•The noise is not sufficient to obscure the structure in the data if weprocess it right.
•There is a huge amount of structure in the data, but the structure is too complicated to be represented bya simple model.
•
•Themain problem is figuring outhow torepresent the complicated structure in a way that can be learned.
页:
[1]